Transcript Internetworking II - Andrew.cmu.edu

Internetworking II
Organizational Communications
and Technologies
Prithvi Rao
H. John Heinz III School of Public
Policy and Management
Carnegie Mellon University
Objectives

Understand how DNS works

Present a DNS scenario
Naming Hosts

Nameserver is vehicle for mapping a name to a
network



telnet akasha.tic.com vs telnet 192.135.128.129
Network object is passed to transport protocol
interface
Naming evolved with other protocols
History of Naming

Predecessor of Internet was ARPANET




Most important resource was IP address
Used naming authority to assign IP addresses
Most hosts had single network interfaces; hostname were
synonymous to interface
Central registry to maintained names and
corresponding IP addresses



Administrator received a host and IP address for each new machine
to be added to network
IP address known when network was established
Name collision avoided by searching a host file
History of Naming

Host files were copied to each machine


Unix systems consisted of /etc/hosts file
Operating systems supported lookup using library
functions




gethostbyname() and gethostbyaddr()
Worked well for small number of hosts (100s)
Other operating systems used similar mechanisms but basically the
same
Worked well because relatively few requests and table size
relatively small
History of Naming

Exponential growth of the internet made static host
table impractical



Load on servers hosting registry introduced delays in access
Names had to be unique to avoid name clashes
Solution to support growing internet was Domain
Name System (DNS)
Domain Name System

Internet’s official naming system




Distributed naming system
Database is scattered across many hosts
Maintained by many organizations (each has a small part)
Defines resource named and protocols used to
communicate between nameservers that maintain
the database
Domain Name System

Delegation



Dynamic Distribution



Naming is delegated leaving central registry to register only naming
authorities
Every host is not named by central authority
Name lookup is dynamically distributed
Site administrators did not have to copy host files
Redundancy


Lookup algorithms were redundant; no single server
Reliability was improved
Domain Name System

Extensibility

Not necessarily restricted to IP addresses
Delegation

Defines a name space that is a tree structure




Each node owned by single authority
Child nodes can be created
Each child node must have a unique name
Domain is any node and its descendant nodes


Domain name uniquely indentifies single node within domain
Node names are written with separated period
Delegation
root
com
edu
cmu
tic
andrew
unix5
akasha
org
kiwilabs
nz …….
co
ac
Delegation


Children of root are “top-level domains”
Domain name that traverses from node to root is
called a Fully Qualified Domain Name (FQDN)




Always ends with a period cs.edu(.)
Practically the period is dropped cs.edu
Some applications (mail) do not permit the appending of a period
Domain name traversing part of node is called a
Relative Domain Name
Dynamic Distribution

Descendants of a domain called subdomains



Naming authority can assign subdomain names
arbitrarily



kiwilabs.com has authority for all names under kiwilabs.com
Grant of authority is given when new subdomain is registered
Child node must be unique
ux4.sp.cs.cmu.edu?
Hierarchy is broader than deeper
Extensibility


Name gives resource a convenient reference; name is
mapped to resource
Can map DNS name to other resources

DNS uses a typed resource record to identify resource being named
<domain-name ttl IN resource_type resource_value)

domain_name is the FQDN for the resource that is
key to identifying resource
Extensibility

ttl is the time to live value



Time that the resource record can be cached before being
discarded
Field is decremented every second and resource is discarded when
ttl reaches zero
IN identifies resource as belonging to TCP/IP or
INternet protocol
Extensibility

resource_type is a unique identifier for type of
resource named



During lookup resource_type is used to distinguish between
resource records mqpped to the domain name
resource_value is value of resource. Can be single
value (IP address) or record with multiple values
DNS has standard set of resource record types
Resource Types

IP addresses
domain_name A ip_addresses
Example
ticmac.tic.com A 192.135.128.131 and A is the record
type corresponding to IP addresses
maps domain name ticmac.tic.com to
192.135.128.131
IP Address

Multi-homed host or router has an A record for each
network interface
router.tic.com A 192.135.128.1
router.tic.com A 193.1.1.1
This illustrates mapping of name router.tic.com to
two IP addresses.
Machine has two interface cards
Host Information

HINFO record indentifies and operating system of
host with given domain name
domain_name HINFO hardware os
Example
akasha.tic.com HINFO Sun SunOs
Alias

Alias is CNAME record associating domain name with
another domain name
domain_name CNAME canonical_name
Example
mac.tic.com CNAME ticmac.tic.com says that name
mac.tic.com is alias for ticmac.tic.com
DNS Operational Architecture
Query or reply
Server
To/from another server
Query or reply
reply
query
Resolver
library
function return
function call
Application
DNS Query Format
header
question
answer
authority
additional
DNS Operational Architecture

question contains the target domain name and the
type and class of query



Can match resource record type or be wildcarded to ask for any
resource
answer is completed by nameserver that replies to
query
authority can name other authority that can answer
query
DNS Operational Architecture

additional completed by nameserver and assists
client with needed information
DNS Operational Steps


Application sends DNS query to nameserver and
waits for response from resolver
Resolver generates query and and transmits it to
nameserver and handles response and retransmits a
query request
Examples of API for DNS
gethostbyname() and gethostbyservice()
DNS Zones
root
com
edu
cmu
tic
andrew
unix5
akasha
org
kiwilabs
nz …….
co
ac
DNS Zones

Each DNS zone has its own zone database


Primary name-server exists for each zone and maintains an up-todate copy of zone database
Copies maintained in secondary nameservers (reliability)
DNS Scenario
1) Query from machine able.widget.com is sent to nameserver
on ns.widget.com for the IP address for the domain name
baker.austin.tic.com: step 1
2) ns.widget.com has no cached resource records for
baker.austin.tic.com so the nameserver tries to find an
NS record for the parent domain austin.tic.com
3) Finding no cached records for that domain it attempts to
find an NS record for the tic.com domain. It looks for the
com domain without success. It forwards original query to
a root nameserver: step 2
DNS Scenario
4) Root nameserver repeats step 3 and finds an NS record
for the com server and passes the query to that server
5) Nameserver for com domain once again repeats above
algorithm and finds NS record and associated A record for
the domain tic.com and returns information to nameserver
on ns.widget.com: step 4
6) Information is cached on ns.widget.com (NS and A
records) and sends original query to server for tic.com.
Second server for that domain is contacted if timeout
occurs: step 5
DNS Scenario
7) Server for tic.com receiving query forwards it to server for
austin.tic.com domain: step 6
8) Destination server has answer desired by original node
(baker.austin.tic.com) and returns answer to tic.com (7)
which then sends answer to ns.widget.com (8) which in turn
returns answer to able.widget.com (9) and this machine
caches answer for later use
Query Example
1
able.widget.com
2
ns.widget.com
rootserver
9
8
5
4
3
6
ns.austin.tic.com
akasha.tic.com
7
comserver
Summary

Presented a brief history of domains and host naming

Examined the use of resource records

Presented DNS query example